CN113095881A

CN113095881A - Method and system for determining pushing elements based on artificial intelligence and big data center

Info

Publication number: CN113095881A
Application number: CN202110414867.1A
Authority: CN
Inventors: 陈炜炜
Original assignee: Individual
Current assignee: Individual
Priority date: 2020-11-11
Filing date: 2020-11-11
Publication date: 2021-07-09
Also published as: CN112308627B; CN113095882A; CN112308627A

Abstract

The embodiment of the application provides a pushing element determining method and system based on artificial intelligence and a big data center, wherein advertisement access data of an advertisement access party is subjected to feature extraction to obtain pushing key element prediction information, and feature information corresponding to historical pushing element information and advertisement access data feature information generated in the feature extraction process are processed to obtain reference pushing element feature information; and finally, determining a target push element set according to the predicted push element characteristic information corresponding to the push key element prediction information and the reference push element characteristic information. Therefore, the characteristics in the pushing element set can be accurately predicted according to the advertisement access data related to the advertisement access party, the pushing element set corresponding to the latest advertisement access record advertisement access process information is generated, the prediction accuracy of the characteristics of the current advertisement access party is improved, the accuracy of the pushing element set is improved, and the matching degree of information pushing is improved.

Description

Method and system for determining pushing elements based on artificial intelligence and big data center

The application is a divisional application of Chinese application with the name of 'pushing element determining method, system and big data center based on artificial intelligence' and is invented and created by application number 202011253979.5, with the application date of 11/2020.

Technical Field

The application relates to the technical field of advertisement pushing based on artificial intelligence, in particular to a pushing element determining method and system based on artificial intelligence and a big data center.

Background

In people's daily life, various advertisements are ubiquitous, the advertisement market is turning from mass marketing to mass marketing, and in the era that products and consumers are subdivided continuously, the limitations of traditional media cannot effectively distinguish target audience groups of products.

For advertising manufacturers, the advertising strategy of the advertising manufacturers involves massive data optimization, and how to improve the accuracy of advertisement data access and thus improve the matching degree of information push is a technical problem to be solved in the field.

Disclosure of Invention

In order to overcome at least the above disadvantages in the prior art, the present application aims to provide a method, a system and a big data center for determining a pushed element based on artificial intelligence, which acquire pushed key element prediction information by performing feature extraction on advertisement access data of an advertisement access party, and process feature information corresponding to historical pushed element information and advertisement access data feature information generated in a feature extraction process to acquire reference pushed element feature information; and finally, determining a target push element set according to the predicted push element characteristic information corresponding to the push key element prediction information and the reference push element characteristic information. Therefore, the characteristics in the pushing element set can be accurately predicted according to the advertisement access data related to the advertisement access party, the pushing element set corresponding to the latest advertisement access record advertisement access process information is generated, the prediction accuracy of the characteristics of the current advertisement access party is improved, the accuracy of the pushing element set is improved, and the matching degree of information pushing is improved.

In a first aspect, the present application provides an advertisement data access method based on a block chain and artificial intelligence, which is applied to a big data center, where the big data center is in communication connection with a plurality of advertisement service terminals, and the method includes:

calling an advertisement push control process for pushing advertisements through a block chain through an advertisement data running process of the big data center, and acquiring advertisement access data related to an advertisement access party based on the advertisement push control process;

extracting characteristics of the advertisement access data to obtain advertisement access data characteristic information, determining push key element prediction information corresponding to a current advertisement access party according to the advertisement access data characteristic information, and extracting the prediction push element characteristic information of the push key element prediction information;

extracting the characteristics of the characteristic information corresponding to the historical pushing element information and the characteristic information of the advertisement access data to obtain reference pushing element characteristic information;

and fusing the predicted pushed element feature information and the reference pushed element feature information to obtain second fused feature information, performing feature analysis on the second fused feature information to obtain current pushed element information, determining a target pushed element set according to the current pushed element information and the historical pushed element information, and pushing corresponding advertisement data to the advertisement service after accessing advertisement data based on the target pushed element set.

In a possible implementation manner of the first aspect, the performing feature extraction on the advertisement access data to obtain advertisement access data feature information includes:

extracting characteristics of various advertisement access track data in the advertisement access data to obtain a plurality of track characteristic information;

fusing the track characteristic information to obtain first fused characteristic information;

and classifying the first fusion characteristic information to acquire the advertisement access data characteristic information.

In one possible implementation manner of the first aspect, the advertisement access data includes: historical advertisement access records, and advertisement access label information, advertisement access interaction information and access configuration information of advertisement access parties corresponding to the information of each advertisement access process in the historical advertisement access records;

the extracting the characteristics of various advertisement access track data in the advertisement access data to obtain a plurality of track characteristic information comprises the following steps:

extracting characteristics of each advertisement access process information in the historical advertisement access records to obtain first track characteristic information corresponding to each advertisement access process information;

extracting the characteristics of the advertisement access interaction information to obtain advertisement access interaction characteristics, and classifying the advertisement access interaction characteristics to obtain second track characteristic information;

indexing in an advertisement access tag lookup table according to the advertisement access tag information to acquire third track characteristic information;

and indexing in an access configuration lookup table according to the access configuration information to acquire fourth track characteristic information.

In a possible implementation manner of the first aspect, the determining, according to the advertisement access data feature information, push key element prediction information corresponding to a current advertisement access party includes:

classifying the advertisement access data characteristic information to obtain an advertisement access data classification result;

and collecting the pushing key elements included by each advertisement access classification label in the advertisement access data classification result to obtain pushing key element prediction information corresponding to the current advertisement access party.

In a possible implementation manner of the first aspect, the extracting features of the feature information corresponding to the historical pushed element information and the advertisement access data feature information to obtain reference pushed element feature information includes:

extracting a feature vector of the historical pushing element information to obtain a historical pushing element vector, and extracting the feature of the historical pushing element vector to obtain the historical pushing element feature information;

extracting the characteristics of the historical pushed element characteristic information and the advertisement access data characteristic information, and classifying the characteristic information after characteristic extraction to obtain the reference pushed element characteristic information;

wherein the obtaining of the second fusion feature information further comprises:

and fusing the predicted pushed element feature information, the advertisement access configuration feature corresponding to the current advertisement access party and the reference pushed element feature information to obtain second fused feature information.

In a possible implementation manner of the first aspect, the step of performing feature analysis on the second fused feature information to obtain current pushed element information, and determining a target pushed element set according to the current pushed element information and the historical pushed element information includes:

acquiring a pushed content service represented by each pushed element in the second fusion characteristic information, acquiring content service characteristics of the pushed content service and source characteristic information corresponding to the previous n historical scheduling advertisement access records, wherein n is a positive integer;

acquiring advertisement source configuration characteristics of a current advertisement source in the source characteristic information, and performing decision tree mining on the content service characteristics, the source characteristic information corresponding to the previous n historical scheduling advertisement access records and the advertisement source configuration characteristics to obtain source characteristic information corresponding to the current advertisement source;

performing decision tree mining on the source characteristic information and the content service characteristic corresponding to the current advertisement source to obtain current pushed element information corresponding to the current advertisement source;

and determining the current pushed element information and the historical pushed element information as a target pushed element set.

In a possible implementation manner of the first aspect, the performing decision tree mining on the content service feature, the source feature information corresponding to the previous n historical scheduling advertisement access records, and the advertisement source configuration feature to obtain the source feature information corresponding to the current advertisement source includes:

acquiring first source characteristic information corresponding to an ith historical scheduling advertisement access record, wherein i is a positive integer and the initial value of i is 1;

performing decision tree mining on the content service characteristics, the first source characteristic information and the advertisement source configuration characteristics, and outputting second source characteristic information corresponding to the (i + 1) th historical scheduling advertisement access record;

repeating the step of outputting the second source characteristic information, and determining the second source characteristic information corresponding to the (n + 1) th history scheduling advertisement access record as the source characteristic information corresponding to the current advertisement source

The performing decision tree mining on the content service features, the first source feature information and the advertisement source configuration features and outputting second source feature information corresponding to an i +1 th historical scheduling advertisement access record includes:

calling a jth output unit to perform decision tree mining on the content service features, the first source feature information and the advertisement source configuration features, and outputting first feature output distribution information, wherein the first feature output distribution information is distribution information corresponding to the (i + 1) th history scheduling advertisement access record;

performing probability estimation processing on the first characteristic output distribution information output by the jth output unit to obtain a first probability classification estimation distribution;

performing probability estimation processing on the first probability classification estimation distribution and the content service features to obtain a second probability classification estimation distribution;

performing probability estimation processing on the second probability classification estimation distribution and the first source characteristic information corresponding to the ith historical scheduling advertisement access record to obtain a third probability classification estimation distribution;

performing probability estimation processing on the third probability classification estimation distribution to obtain second characteristic output distribution information output by the j +1 th output unit, wherein the second characteristic output distribution information is distribution information corresponding to the i +1 th historical scheduling advertisement access record, j +1 is not less than k, j is a positive integer and the initial value of j is 1;

and repeating the step of outputting the second characteristic output distribution information, and determining the second characteristic output distribution information output by the last output unit as second source characteristic information corresponding to the (i + 1) th historical scheduling advertisement access record.

In a possible implementation manner of the first aspect, the performing decision tree mining on the source feature information and the content service feature corresponding to the current advertisement source to obtain current pushed element information corresponding to the current advertisement source includes:

acquiring the push element characteristics of the push elements which are output in the actual push element information corresponding to the current advertisement source;

performing decision tree mining on the source characteristic information, the content service characteristic and the pushing element characteristic corresponding to the current advertisement source, and outputting the current pushing element information corresponding to the current advertisement source;

the performing decision tree mining on the source feature information, the content service feature and the push element feature corresponding to the current advertisement source and outputting the current push element information corresponding to the current advertisement source includes:

calling an mth mining unit to perform decision tree mining on the source feature information, the content service feature and the pushing element feature corresponding to the current advertisement source, and outputting third feature output distribution information corresponding to the current advertisement source;

performing intermediate decoding processing on the third feature output distribution information output by the mth mining unit to obtain a first probability classification estimation distribution;

performing probability classification estimation processing on the first probability classification estimation distribution, the content service features and source feature information corresponding to the current advertisement source to obtain a second probability classification estimation distribution;

performing feature extraction on the second probability classification estimation distribution to obtain fourth feature output distribution information corresponding to the current advertisement source output by the (m + 1) th mining unit, wherein m +1 is not more than t, m is a positive integer and the initial value of m is 1;

and repeating the step of outputting the fourth characteristic output distribution information, and determining the push element output by the last mining unit as the current push element information corresponding to the current advertisement source.

In a possible implementation manner of the first aspect, the step of invoking, by the advertisement data running process of the big data center, an advertisement push control process that performs advertisement push through a block chain includes:

obtaining visitor interest data corresponding to advertisement push feedback information of a target advertisement push task, and searching crowd interest points based on the visitor interest data;

acquiring crowd interest point information corresponding to each crowd interest point in a preset number of crowd interest points, wherein the crowd interest point information comprises crowd interest point coverage labels, interest point advertisement labels and access attribute information, the crowd interest point coverage labels are used for the crowd labels covered by the crowd interest points, the interest point advertisement labels are used for indicating advertisement attention contents of the crowd interest points, and the access attribute information is used for indicating advertisement access conditions in the crowd interest points;

generating advertisement pushing optimization information corresponding to each crowd interest point according to the crowd interest point information corresponding to each crowd interest point, wherein the advertisement pushing optimization information is used for carrying out advertisement putting strategy node optimization processing or advertisement putting strategy node strengthening processing on advertisement strategy starting information, and the advertisement pushing optimization information and the crowd interest points have one-to-one correspondence relation;

processing advertisement strategy starting information corresponding to each crowd interest point by adopting the advertisement push optimization information corresponding to each crowd interest point to obtain advertisement strategy confirmation information corresponding to each crowd interest point, and generating an advertisement push control process for pushing advertisements through a block chain according to the advertisement strategy confirmation information corresponding to each crowd interest point, wherein the advertisement push optimization information, the advertisement strategy starting information and the advertisement strategy confirmation information have a one-to-one correspondence relationship;

and calling the advertisement push control process through the advertisement data running process of the big data center.

In a second aspect, an embodiment of the present application further provides an advertisement data access apparatus based on a block chain and artificial intelligence, which is applied to a big data center, where the big data center is in communication connection with a plurality of advertisement service terminals, and the apparatus includes:

the calling module is used for calling an advertisement pushing control process for pushing advertisements through a block chain through an advertisement data running process of the big data center, and acquiring advertisement access data related to an advertisement access party based on the advertisement pushing control process;

the first extraction module is used for extracting the characteristics of the advertisement access data to obtain advertisement access data characteristic information, determining push key element prediction information corresponding to a current advertisement access party according to the advertisement access data characteristic information, and extracting the predicted push key element characteristic information of the push key element prediction information;

the second extraction module is used for extracting the characteristics of the characteristic information corresponding to the historical pushed element information and the characteristic information of the advertisement access data to obtain reference pushed element characteristic information;

the determining module is configured to fuse the predicted pushed element feature information and the reference pushed element feature information to obtain second fused feature information, perform feature analysis on the second fused feature information to obtain current pushed element information, determine a target pushed element set according to the current pushed element information and the historical pushed element information, and push corresponding advertisement data to the advertisement service after performing advertisement data access based on the target pushed element set.

In a third aspect, an embodiment of the present application further provides an advertisement data access system based on a block chain and artificial intelligence, where the advertisement data access system based on the block chain and artificial intelligence includes a big data center and a plurality of advertisement service terminals in communication connection with the big data center;

the big data center is used for:

In a fourth aspect, an embodiment of the present application further provides a big data center, where the big data center includes a processor, a machine-readable storage medium, and a network interface, where the machine-readable storage medium, the network interface, and the processor are connected through a bus system, the network interface is configured to be communicatively connected to at least one advertisement service terminal, the machine-readable storage medium is configured to store a program, an instruction, or a code, and the processor is configured to execute the program, the instruction, or the code in the machine-readable storage medium to perform the method for accessing advertisement data based on a blockchain and artificial intelligence in the first aspect or any one of the possible implementations of the first aspect.

In a fifth aspect, an embodiment of the present application provides a computer-readable storage medium, where instructions are stored, and when executed, cause a computer to perform a block chain and artificial intelligence based advertisement data access method in the first aspect or any one of the possible implementation manners of the first aspect.

Based on any one of the above aspects, the method obtains the pushed key element prediction information by performing feature extraction on advertisement access data of an advertisement access party, and obtains reference pushed element feature information by processing feature information corresponding to historical pushed element information and advertisement access data feature information generated in a feature extraction process; and finally, determining a target push element set according to the predicted push element characteristic information corresponding to the push key element prediction information and the reference push element characteristic information. Therefore, the characteristics in the pushing element set can be accurately predicted according to the advertisement access data related to the advertisement access party, the pushing element set corresponding to the latest advertisement access record advertisement access process information is generated, the prediction accuracy of the characteristics of the current advertisement access party is improved, the accuracy of the pushing element set is improved, and the matching degree of information pushing is improved.

Drawings

In order to more clearly illustrate the technical solutions of the embodiments of the present application, the drawings that need to be called in the embodiments are briefly described below, it should be understood that the following drawings only illustrate some embodiments of the present application and therefore should not be considered as limiting the scope, and for those skilled in the art, other related drawings can be obtained according to the drawings without inventive efforts.

Fig. 1 is a schematic view of an application scenario of an advertisement data access system based on a blockchain and artificial intelligence provided in an embodiment of the present application;

fig. 2 is a schematic flowchart of an advertisement data access method based on a blockchain and artificial intelligence according to an embodiment of the present application;

fig. 3 is a schematic functional block diagram of an advertisement data access device based on a blockchain and artificial intelligence according to an embodiment of the present application;

fig. 4 is a schematic block diagram of structural components of a big data center for implementing the above-described advertisement data access method based on blockchains and artificial intelligence according to an embodiment of the present disclosure.

Detailed Description

The present application will now be described in detail with reference to the drawings, and the specific operations in the method embodiments may also be applied to the apparatus embodiments or the system embodiments.

Fig. 1 is an interaction diagram of an advertisement data access system 10 based on blockchain and artificial intelligence according to an embodiment of the present application. The blockchain and artificial intelligence based advertising data access system 10 may include a big data center 100 and an advertising service terminal 200 communicatively connected with the big data center 100. The blockchain and artificial intelligence based advertising data access system 10 shown in fig. 1 is merely one possible example, and in other possible embodiments, the blockchain and artificial intelligence based advertising data access system 10 may also include only some of the components shown in fig. 1 or may also include other components.

In this embodiment, the big data center 100 and the advertisement service terminal 200 in the advertisement data access system 10 based on the blockchain and the artificial intelligence may cooperatively perform the advertisement data access method based on the blockchain and the artificial intelligence described in the following method embodiment, and the detailed description of the following method embodiment may be referred to in the execution step portions of the big data center 100 and the advertisement service terminal 200.

To solve the technical problem in the foregoing background, fig. 2 is a schematic flowchart of an advertisement data access method based on a blockchain and artificial intelligence according to an embodiment of the present application, where the advertisement data access method based on a blockchain and artificial intelligence according to the present embodiment may be executed by the big data center 100 shown in fig. 1, and the advertisement data access method based on a blockchain and artificial intelligence is described in detail below.

Step S110, an advertisement push control process for pushing advertisements through a block chain is called through an advertisement data running process of the big data center, and advertisement access data related to an advertisement access party is obtained based on the advertisement push control process.

For example, an advertisement access party to be accessed can be determined based on an advertisement push control process, and then relevant advertisement access data of the advertisement access party is obtained from an advertisement server correspondingly configured by the advertisement access party, and the advertisement access data can be used for characterizing the situation of push key elements involved in advertisement access of the advertisement access party.

Step S120, extracting the characteristics of the advertisement access data to obtain the characteristic information of the advertisement access data, determining the prediction information of the push key elements corresponding to the current advertisement access party according to the characteristic information of the advertisement access data, and extracting the prediction push element characteristic information of the prediction information of the push key elements.

In this embodiment, the push key element may refer to a push item (for example, a push item of a certain order service), a push group (for example, a group associated with a certain order service), and the like involved in the process of performing advertisement data access.

Step S130, performing feature extraction on feature information corresponding to the historical pushed element information and advertisement access data feature information to obtain reference pushed element feature information.

Step S140, fusing the predicted pushed element feature information and the reference pushed element feature information to obtain second fused feature information, performing feature analysis on the second fused feature information to obtain current pushed element information, determining a target pushed element set according to the current pushed element information and the historical pushed element information, and pushing corresponding advertisement data to an advertisement service after accessing advertisement data based on the target pushed element set.

In this embodiment, after the target pushed element set is obtained, the advertisement data can be accessed based on each pushed element in the target pushed element set, and then the corresponding advertisement data is pushed to the advertisement service.

Based on the above steps, the embodiment performs feature extraction on advertisement access data of an advertisement access party to obtain push key element prediction information, and processes feature information corresponding to historical push element information and advertisement access data feature information generated in the feature extraction process to obtain reference push element feature information; and finally, determining a target push element set according to the predicted push element characteristic information corresponding to the push key element prediction information and the reference push element characteristic information. Therefore, the characteristics in the pushing element set can be accurately predicted according to the advertisement access data related to the advertisement access party, the pushing element set corresponding to the latest advertisement access record advertisement access process information is generated, the prediction accuracy of the characteristics of the current advertisement access party is improved, the accuracy of the pushing element set is improved, and the matching degree of information pushing is improved.

In one possible implementation manner, for step S120, in the process of performing feature extraction on the advertisement access data to obtain advertisement access data feature information, the following exemplary sub-steps may be implemented.

And a substep S121, performing feature extraction on various advertisement access track data in the advertisement access data to obtain a plurality of track feature information.

And a substep S122, fusing the track feature information to obtain first fused feature information.

And a substep S123 of classifying the first fusion characteristic information to obtain advertisement access data characteristic information.

For example, the advertisement access data may include: historical advertisement access records, advertisement access label information, advertisement access interaction information and access configuration information of advertisement access parties corresponding to the advertisement access process information in the historical advertisement access records.

In sub-step S121, feature extraction may be performed on each advertisement access process information in the historical advertisement access record to obtain first track feature information corresponding to each advertisement access process information.

And, feature extraction may be performed on the advertisement access interaction information to obtain advertisement access interaction features, and the advertisement access interaction features may be classified to obtain second trajectory feature information.

Meanwhile, indexing can be performed in the advertisement access tag lookup table according to the advertisement access tag information to acquire third track characteristic information.

Finally, the fourth track characteristic information may be obtained by indexing in the access configuration lookup table according to the access configuration information.

Thus, the first track characteristic information, the second track characteristic information, the third track characteristic information, and the fourth track characteristic information can be summarized as a plurality of track characteristic information.

In one possible implementation manner, for step S120, in determining the push key element prediction information corresponding to the current advertisement access party according to the advertisement access data feature information, the following exemplary sub-steps may be implemented.

And a substep S124, classifying the advertisement access data characteristic information to obtain an advertisement access data classification result.

And a substep S125, collecting the pushing key elements included by each advertisement access classification label in the advertisement access data classification result to obtain pushing key element prediction information corresponding to the current advertisement access party.

In this way, the effective information component of the configuration information of each predicted push key element in the push key element prediction information can be used as the reference push element feature information in advance.

In one possible implementation manner, for step S130, in the process of performing feature extraction on feature information corresponding to the historical push element information and advertisement access data feature information to obtain reference push element feature information, the following exemplary sub-steps may be implemented.

The substep S131 extracts a feature vector of the history pushed element information to obtain a history pushed element vector, and extracts a feature of the history pushed element vector to obtain history pushed element feature information.

And a substep S132 of extracting the characteristics of the historical pushed element characteristic information and the advertisement access data characteristic information, and classifying the characteristic information after characteristic extraction to obtain reference pushed element characteristic information.

In step S140, in the process of obtaining the second fusion characteristic information, the predicted push feature information, the advertisement access configuration feature corresponding to the current advertisement access party, and the reference push feature information may be fused to obtain the second fusion characteristic information.

The specific fusion mode may refer to a data fusion algorithm in the prior art, for example, a data in and data out (DAI-DAO) fusion algorithm, which is not described herein again.

In one possible implementation manner, still referring to step S140, in performing feature analysis on the second fused feature information to obtain current pushed element information, and determining a target pushed element set according to the current pushed element information and the historical pushed element information, the following exemplary sub-steps may be implemented.

And a substep S141, obtaining the push content service represented by each push element in the second fusion characteristic information, and obtaining the content service characteristic of the push content service and the source characteristic information corresponding to the previous n historical scheduling advertisement access records, wherein n is a positive integer.

And a substep S142, obtaining the advertisement source configuration characteristics of the current advertisement source in the source characteristic information, and performing decision tree mining on the content service characteristics, the source characteristic information corresponding to the previous n historical scheduling advertisement access records and the advertisement source configuration characteristics to obtain the source characteristic information corresponding to the current advertisement source.

For example, first source characteristic information corresponding to the ith historically scheduled advertisement access record may be obtained, where i is a positive integer and the starting value of i is 1.

And then, performing decision tree mining on the content service characteristics, the first source characteristic information and the advertisement source configuration characteristics, and outputting second source characteristic information corresponding to the (i + 1) th historical scheduling advertisement access record.

And repeating the step of outputting the second source characteristic information, and determining the second source characteristic information corresponding to the (n + 1) th historical scheduling advertisement access record as the source characteristic information corresponding to the current advertisement source.

For example, the jth output unit may be invoked to perform decision tree mining on the content service feature, the first source feature information, and the advertisement source configuration feature, and output first feature output distribution information, where the first feature output distribution information is distribution information corresponding to the i +1 th historically scheduled advertisement access record.

Next, probability estimation processing may be performed on the first feature output distribution information output by the jth output unit to obtain a first probability classification estimation distribution.

Then, the probability estimation process can be performed on the first probability classification estimation distribution and the content service feature to obtain a second probability classification estimation distribution.

Then, probability estimation processing can be performed on the second probability classification estimation distribution and the first source characteristic information corresponding to the ith historical scheduling advertisement access record to obtain a third probability classification estimation distribution.

Then, probability estimation processing may be performed on the third probability classification estimation distribution to obtain second feature output distribution information output by the j +1 th output unit, where the second feature output distribution information is distribution information corresponding to the i +1 th historical scheduling advertisement access record, j +1 is not less than k, j is a positive integer, and the initial value of j is 1.

In this way, the step of outputting the second feature output distribution information is repeated, and the second feature output distribution information output by the last output unit is determined as the second source feature information corresponding to the i +1 th historically scheduled advertisement access record.

And a substep S143, performing decision tree mining processing on the source characteristic information and the content service characteristic corresponding to the current advertisement source to obtain the current pushed element information corresponding to the current advertisement source.

For example, the substep S143 may be implemented by the following exemplary embodiments.

(1) And acquiring the push element characteristics of the push elements which are output in the actual push element information corresponding to the current advertisement source.

(2) And performing decision tree mining processing on the source characteristic information, the content service characteristic and the pushing element characteristic corresponding to the current advertisement source, and outputting the current pushing element information corresponding to the current advertisement source.

(3) Performing decision tree mining processing on source characteristic information, content service characteristics and pushing element characteristics corresponding to the current advertisement source, and outputting current pushing element information corresponding to the current advertisement source, wherein the decision tree mining processing comprises the following steps:

(4) and calling the mth mining unit to perform decision tree mining on the source feature information, the content service feature and the pushing element feature corresponding to the current advertisement source, and outputting third feature output distribution information corresponding to the current advertisement source.

(5) And performing intermediate decoding processing on the third characteristic output distribution information output by the mth mining unit to obtain a first probability classification estimation distribution.

(6) And carrying out probability classification estimation processing on the first probability classification estimation distribution, the content service characteristics and the source characteristic information corresponding to the current advertisement source to obtain a second probability classification estimation distribution.

(7) And performing feature extraction on the second probability classification estimation distribution to obtain fourth feature output distribution information corresponding to the current advertisement source output by the (m + 1) th mining unit, wherein m +1 is not more than t, m is a positive integer and the initial value of m is 1.

In this way, the step of outputting the fourth feature output distribution information may be repeated, and the push element output by the last mining unit may be determined as the current push element information corresponding to the current advertisement source.

And a substep S144, determining the current pushed element information and the historical pushed element information as a target pushed element set.

Based on the substeps, the pushed content service and the previous n historical scheduling advertisement access records related to the pushed content service are processed, so that the actual pushed element information corresponding to the current advertisement source can be better output in the advertisement access process according to various types of information, the pushing accuracy is improved, the consistency of the output pushed element information, the advertisement source and the pushed content service is ensured, and the advertisement access effect is improved.

In one possible implementation manner, for step S110, in the process of invoking an advertisement push control process for advertisement push through a blockchain through an advertisement data running process of the big data center, the following exemplary sub-steps may be implemented.

And a substep S111 of obtaining visitor interest data corresponding to the advertisement push feedback information of the target advertisement push task and searching crowd interest points based on the visitor interest data.

And a substep S112, obtaining the crowd interest point information corresponding to each crowd interest point in the preset number of crowd interest points.

In this embodiment, the crowd interest point information may specifically include a crowd interest point coverage tag, an interest point advertisement tag, and access attribute information. For example, the crowd interest point coverage tag may be used for a crowd tag covered by a crowd interest point (for example, a crowd tag using a certain APP, a crowd tag scanning a certain product two-dimensional code, or the like), the interest point advertisement tag is used for indicating advertisement attention content of the crowd interest point, the advertisement attention content may be a user click content area corresponding to an advertisement playing stream, and the access attribute information is used for indicating an advertisement access condition in the crowd interest point, for example, a click access condition, a collection access condition, a sharing access condition, or the like.

And a substep S113, generating advertisement push optimization information corresponding to each crowd interest point according to the crowd interest point information corresponding to each crowd interest point.

In this embodiment, the advertisement push optimization information is used to perform advertisement delivery policy node optimization processing or advertisement delivery policy node enhancement processing on the advertisement policy initiation information, and the advertisement push optimization information and the crowd interest points have a one-to-one correspondence relationship. The advertisement putting strategy node optimization processing on the advertisement strategy initial information can mean that the advertisement strategy initial information has advertisement putting strategy nodes and needs adaptive optimization, so that the advertisement strategy initial information is adaptive to the crowd interest point information corresponding to the current crowd interest point. The advertisement putting strategy node strengthening processing on the advertisement strategy initial information can mean that the advertisement strategy initial information has the advertisement putting strategy node which needs to be strengthened in a self-adaptive mode, and the putting strength of the advertisement putting strategy node can be strengthened.

And a substep S114 of processing the advertisement strategy initial information corresponding to each crowd interest point by adopting the advertisement push optimization information corresponding to each crowd interest point to obtain the advertisement strategy confirmation information corresponding to each crowd interest point, and generating an advertisement push control process for pushing the advertisement through the block chain according to the advertisement strategy confirmation information corresponding to each crowd interest point.

The advertisement pushing optimization information, the advertisement strategy starting information and the advertisement strategy confirmation information have a one-to-one correspondence relationship. In this way, by generating the final advertisement push control process, the configuration information for performing crowd interest point retrieval on each advertisement attention content can be regenerated based on the final advertisement push control process, thereby performing closed-loop feedback control.

Based on the above steps, the present embodiment can generate advertisement push optimization information corresponding to each crowd interest point according to the crowd interest point information corresponding to each crowd interest point, and then process the advertisement policy start information corresponding to each crowd interest point to obtain the advertisement policy confirmation information corresponding to each crowd interest point, thereby generating an advertisement push control process for pushing advertisements through a block chain. By adopting the mode, the advertisement strategy information from different dimensions is processed in parallel based on the plurality of crowd interest points, and then under the multi-advertisement strategy scene, the advertisement putting strategy node strengthening or the advertisement putting strategy node optimizing can be carried out on the advertisement strategy information on different dimensions through the advertisement pushing optimizing information, so that the advertisement strategy information can be optimized in real time, the accuracy of configuration information in the advertisement strategy detection process is improved, and the subsequent advertisement strategy processing effect is favorably improved.

For example, in a possible implementation manner, for step S112, in the process of acquiring the crowd point-of-interest information corresponding to each crowd point-of-interest in the preset number of crowd points-of-interest, the following exemplary sub-steps may be implemented, which are described in detail below.

And a substep S1121, detecting each crowd interest point in a preset number of crowd interest points to obtain an advertisement access detection result corresponding to each crowd interest point.

In the sub-step S1122, access attribute information corresponding to each of the crowd interest points is determined according to the advertisement access detection result corresponding to each of the crowd interest points.

And a substep S1123 of determining access migration relationship information corresponding to each crowd interest point according to the advertisement access detection result corresponding to each crowd interest point.

For example, for any one of a preset number of crowd interest points, if the advertisement access detection result indicates that access object information with access node migration exists in the crowd interest points, the information of the access node migration process of the access object information in the crowd interest points is determined as access migration relationship information corresponding to each crowd interest point.

In sub-step S1124, a crowd interest point coverage tag corresponding to each crowd interest point and an interest point advertisement tag corresponding to each crowd interest point are obtained.

And S1125, generating the crowd interest point information corresponding to each crowd interest point according to the access attribute information corresponding to each crowd interest point, the access migration relationship information corresponding to each crowd interest point, the crowd interest point coverage label corresponding to each crowd interest point and the interest point advertisement label corresponding to each crowd interest point.

Further, on this basis, as a possible implementation manner, in the process of generating the advertisement push optimization information corresponding to each crowd interest point according to the crowd interest point information corresponding to each crowd interest point, aiming at the step S113, the following exemplary sub-steps may be implemented, which are described in detail as follows.

And a substep S1131, obtaining the relationship entity information corresponding to the interest point advertisement tag from the advertisement push strategy demand information corresponding to the pre-configured crowd interest point coverage tag.

And a substep S1132, performing feature extraction on the relationship entity information according to the access attribute information to obtain push optimized context information in which the relationship entity information is respectively matched with the access attribute information.

And a substep S1133, determining global advertisement push optimization information based on the push optimization scenario information of the relationship entity information.

And a substep S1134, determining access migration relationship entity information in the relationship entity information according to the access migration relationship information, and determining push optimization scenario information corresponding to the access migration relationship entity information.

And a substep S1135, fusing the global advertisement push optimization information and the push optimization context information corresponding to the access migration relationship entity information to obtain advertisement push optimization information corresponding to each crowd interest point.

In this embodiment, the advertisement push optimization information includes an advertisement delivery policy node that needs to perform advertisement delivery policy node optimization processing or advertisement delivery policy node enhancement processing on the advertisement policy initiation information.

The access attribute information comprises an advertisement access thermodynamic diagram, the advertisement access thermodynamic diagram comprises a plurality of thermal units and thermal keyword characteristics connected between the two thermal units, the thermal keyword characteristics comprise thermal keyword labels and thermal keyword word vectors of the thermal keyword characteristics, and the thermal units comprise relationship entity thermal nodes and a hot spot map.

On this basis, for sub-step S1132, it may be implemented by the following exemplary embodiments.

(1) And determining a relationship entity thermodynamic node corresponding to the relationship entity information in the advertisement access thermodynamic diagram.

(2) And determining the thermal keyword parameter of the relationship entity thermal node and optimizing the thermal keyword parameter in a plurality of thermal units of the advertisement access thermodynamic diagram according to the thermal keyword label.

(3) And calculating a first push optimization scene generated by the released thermal keyword parameters to the relationship entity information according to the thermal keyword word vectors of the thermal keyword characteristics between the connection relationship entity thermal nodes and the released thermal keyword parameters.

(4) And calculating a second pushing optimization scene generated by the released thermal keyword parameters to the relationship entity information according to the thermal keyword word vector of the thermal keyword characteristics between the connection relationship entity thermal nodes and the released thermal keyword parameters.

(5) And determining the push optimization scenario information of the relationship entity information according to the first push optimization scenario and the second push optimization scenario.

Thus, for step S114, in the process of obtaining the advertisement policy confirmation information corresponding to each crowd point of interest by processing the advertisement policy start information corresponding to each crowd point of interest by using the advertisement push optimization information corresponding to each crowd point of interest, the following exemplary sub-steps can be implemented, which are described in detail below.

And a substep S1141, obtaining a first advertisement putting strategy node which needs to carry out the optimization processing of the advertisement putting strategy node on the advertisement strategy initial information and a second advertisement putting strategy node which needs to carry out the enhancement processing of the advertisement putting strategy node on the advertisement strategy initial information according to the advertisement putting strategy node which needs to carry out the optimization processing of the advertisement putting strategy node or the enhancement processing of the advertisement putting strategy node in the advertisement putting optimization information.

And a substep S1142 of optimizing the first advertisement delivery strategy node according to the corresponding optimization strategy information in the advertisement delivery optimization information and performing reinforcement processing on the second advertisement delivery strategy node according to the corresponding reinforcement strategy information in the advertisement delivery optimization information.

In a possible implementation manner, still regarding step S114, in the process of generating an advertisement push control process for pushing advertisements through a blockchain according to the advertisement policy confirmation information corresponding to each crowd interest point, the process may be implemented by the following exemplary sub-steps, which are described in detail below.

And a substep S1143 of determining an advertisement push preference intention matrix corresponding to each crowd interest point according to the advertisement strategy confirmation information corresponding to each crowd interest point, wherein the advertisement push preference intention matrix is the advertisement push preference intention matrix of the advertisement strategy confirmation information on each advertisement strategy unit.

And a substep S1144 of determining the pushing control information of the advertisement pushing control link file corresponding to each crowd interest point according to the advertisement pushing preference intention matrix corresponding to each crowd interest point.

And a substep S1145 of performing redirection configuration on the current advertisement push control link file based on the push control information of the advertisement push control link file corresponding to each crowd interest point to obtain an advertisement push control process corresponding to each crowd interest point.

For example, the push control associated information of the push control information of the advertisement push control link file corresponding to each crowd interest point for each link jump information in the current advertisement push control link file may be obtained, and the current advertisement push control link file may be redirected and configured based on the push control associated information of each link jump information in the current advertisement push control link file, so as to obtain the advertisement push control process corresponding to each crowd interest point.

In a possible implementation manner, based on the above description, for step S111, in the process of obtaining visitor interest data corresponding to the advertisement push feedback information of the target advertisement push task and performing the search for the crowd interest point based on the visitor interest data, the following exemplary sub-steps may be implemented, and are described in detail as follows.

And a substep S1111, acquiring advertisement control node information corresponding to each advertisement push segment of the advertisement service terminal, and performing advertisement drainage search on advertisement push feedback information of the target advertisement push task based on the advertisement control node information to acquire a corresponding advertisement drainage search segment.

In this embodiment, the advertisement control node information may be used to represent a control node instruction for performing advertisement drainage search interest point retrieval during the playing process of the advertisement playing stream, the target advertisement pushing task may refer to a pushing control service item of the advertisement pushing task data, and the advertisement pushing feedback information may refer to a production monitoring configuration parameter related to the initiated production monitoring request. The advertisement drainage search segment may be a search segment that is obtained from a preconfigured index database for the advertisement push feedback information of the target advertisement push task and matches with the control node instruction retrieved by each advertisement drainage search interest point, and specifically may include drainage behavior obtained by drainage, drainage duration, and the like.

And a substep S1112, obtaining corresponding advertisement diversion skip subgraphs based on the advertisement diversion searching segments, and determining user interest learning information of a plurality of advertisement diversion paths based on the advertisement diversion skip subgraphs.

In this embodiment, the advertisement diversion skip subgraph may include skip record conditions of advertisement diversion processes in each advertisement playing process, the advertisement diversion path may refer to an advertisement diversion path node set formed by each advertisement playing process, and the user interest learning information may refer to a feature extraction condition of user interest in each advertisement playing process in scheduling the advertisement diversion process.

And a substep S1113, respectively inputting the interest learning information of the user into a plurality of interest point retrieval units in the interest point retrieval model, and performing at least one time of advertisement attention content matching through the interest point retrieval units to obtain at least one advertisement attention content.

In this embodiment, at least one matching of the advertisement attention content by the interest point retrieving unit is performed based on the associated big data retrieving partition, and the associated big data retrieving partition is associated with the advertisement attention content extracted by other interest point retrieving units of the plurality of interest point retrieving units. The interest point retrieval unit is obtained based on an artificial intelligence network and corresponding training samples in a training mode, and the training samples comprise user interest learning information samples and corresponding advertisement attention content labels.

And a substep S1114, performing interest point retrieval on the advertisement attention content output by the interest point retrieval units to obtain interest point retrieval content, obtaining visitor interest data of the advertisement diversion skip subgraph under the advertisement push feedback information based on the interest point retrieval content, and performing crowd interest point retrieval based on the visitor interest data of the advertisement diversion skip subgraph under the advertisement push feedback information.

In this embodiment, the visitor interest data of the advertisement drainage skip subgraph under the advertisement push feedback information may be used to represent a retrieval instruction running set of the advertisement drainage skip subgraph in a subsequent crowd interest point retrieval process, that is, a retrieval control instruction for controlling a flow direction relationship of retrieval data nodes in the crowd interest point retrieval process, so as to execute the computer program according to an execution sequence of the retrieval control instructions, thereby performing crowd interest point retrieval.

Based on the above steps, in this embodiment, the user interest learning information of a plurality of advertisement drainage paths is determined based on the advertisement drainage skip subgraph, the user interest learning information is respectively input into a plurality of interest point retrieval units in the interest point retrieval model, each interest point retrieval unit performs at least one advertisement attention content matching to obtain at least one advertisement attention content, and the at least one advertisement attention content matching is performed based on the associated big data retrieval partition, the associated big data retrieval partition is associated with the advertisement attention content extracted by other interest point retrieval units of the plurality of interest point retrieval units, wherein the interest point retrieval unit is obtained based on artificial intelligence network and corresponding training sample training, the training sample includes user interest learning information samples and corresponding advertisement attention content labels, so that the advertisement attention content extracted from different interest point retrieval units can be exchanged and fused at least once, and then, interest point retrieval can be carried out on advertisement attention contents in different levels, so that the representation capability of crowd interest point retrieval is improved by enriching the levels of the advertisement attention contents, and the retrieval pertinence is better.

In a possible implementation manner, on the basis of the above scheme, in this embodiment, one of the interest point retrieving units may further serve as a target interest point retrieving unit, and then a first advertisement attention content extracted by the target interest point retrieving unit and a second advertisement attention content extracted by other interest point retrieving units except the target interest point retrieving unit in the plurality of interest point retrieving units are obtained.

Therefore, when the advertisement drainage path of the second advertisement attention content is not matched with the advertisement drainage path of the first advertisement attention content, the interest degree distribution marking is carried out on the second advertisement attention content, and the advertisement drainage path of the second advertisement attention content after the interest degree distribution marking is the same as the advertisement drainage path of the first advertisement attention content. Therefore, the target interest point retrieval unit can be used for matching the second advertisement attention content marked by the interest degree distribution with the content matching information of the first advertisement attention content.

Wherein the number of the second advertisement attention content is at least two.

On the basis, when second advertisement attention content of the advertisement drainage path, of which the advertisement drainage path does not match the first advertisement attention content, and second advertisement attention content of the advertisement drainage path, of which the advertisement drainage path matches the first advertisement attention content exist at the same time, interest degree distribution marking is performed on the second advertisement attention content of the advertisement drainage path, of which the advertisement drainage path does not match the first advertisement attention content, reverse interest degree distribution marking is performed on the second advertisement attention content of the advertisement drainage path, of which the advertisement drainage path matches the first advertisement attention content, and the advertisement drainage path of the second advertisement attention content after interest degree distribution marking and the advertisement drainage path of the second advertisement attention content after reverse interest degree distribution marking are the same as the advertisement drainage path of the first advertisement attention content.

In this way, the target interest point retrieval unit may perform advertisement attention content matching on the second advertisement attention content marked by the interest degree distribution, the second advertisement attention content marked by the reverse interest degree distribution, and the content matching information of the first advertisement attention content.

For another example, on the basis, when the advertisement drainage path of the second advertisement attention content matches the advertisement drainage path of the first advertisement attention content, the reverse interestingness distribution marking is performed on the second advertisement attention content, and the advertisement drainage path of the second advertisement attention content after the reverse interestingness distribution marking is the same as the advertisement drainage path of the first advertisement attention content. In this way, the target interest point retrieval unit can be used for matching the second advertisement attention content marked by the reverse interestingness distribution with the content matching information of the first advertisement attention content.

For example, in one possible implementation manner, regarding step S1113, in the process of inputting the user interest learning information into a plurality of interest point retrieving units in the interest point retrieving model respectively, and obtaining at least one advertisement attention content by performing at least one advertisement attention content matching through the interest point retrieving units, the following exemplary sub-steps can be implemented, which are described in detail below.

In the substep S11131, the user interest learning information is input to a plurality of interest point retrieval units in the interest point retrieval model, respectively.

And a substep S11132, for one of the interest point retrieval units, matching the advertisement attention content with the corresponding user interest learning information through the interest point retrieval unit, acquiring second advertisement attention content extracted by other interest point retrieval units of the interest point retrieval units after the first advertisement attention content is obtained through the advertisement attention content matching, performing partition service matching with the first advertisement attention content, and continuing to perform advertisement attention content matching based on the partition service matching result so as to alternately perform advertisement attention content matching and partition service matching.

Still further, in step S1112, in the process of determining the user interest learning information of a plurality of advertisement drainage paths based on the advertisement drainage skip subgraph, index matching of the advertisement drainage paths may be performed on the advertisement drainage skip subgraph to obtain the user interest learning information of a plurality of different advertisement drainage paths.

For example, the search information nodes of the advertisement diversion skip subgraph for each advertisement diversion path can be indexed and matched, and all the search information nodes are fused according to the association relationship among the search information nodes, so that the user interest learning information of a plurality of different advertisement diversion paths is obtained.

Thus, in another parallel possible implementation manner, for step S1113, in the process of inputting the user interest learning information into a plurality of interest point retrieving units in the interest point retrieving model respectively, and performing at least one advertisement attention content matching through the interest point retrieving units to obtain at least one advertisement attention content, the following exemplary sub-steps can be implemented, which are described in detail below.

In the substep S11133, the user interest learning information is input to a plurality of interest point retrieval units in the interest point retrieval model, respectively.

The user interest learning information is respectively in one-to-one correspondence with one of the interest point retrieval units.

And a substep S11134 of performing at least one advertisement attention content matching on the corresponding user interest learning information through the interest point retrieval unit. The extracted advertisement drainage path of the advertisement attention content is consistent with the advertisement drainage path of the user interest learning information corresponding to the interest point retrieval unit.

In a parallel possible implementation manner, the user interest learning information at least includes first user interest learning information, second user interest learning information, and third user interest learning information, the interest point retrieval model includes a first interest point retrieval unit, a second interest point retrieval unit, and a third interest point retrieval unit, where the first user interest learning information, the second user interest learning information, and the third user interest learning information respectively correspond to user interest learning information of an interest coverage learning segment, an interest data source segment, and an interest output source segment, and the first interest point retrieval unit, the second interest point retrieval unit, and the third interest point retrieval unit respectively correspond to interest point retrieval units of the interest coverage learning segment, the interest data source segment, and the interest output source segment.

On this basis, still referring to step S1113, in the process of inputting the user interest learning information into the multiple interest point retrieving units in the interest point retrieving model respectively, and performing at least one advertisement attention content matching through the interest point retrieving units to obtain at least one advertisement attention content, the following exemplary sub-steps can be implemented, which are described in detail below.

And a substep S11135, inputting the first user interest learning information into the first interest point retrieval unit to match the advertisement attention content of the first interest point retrieval layer, so as to obtain the advertisement attention content extracted by the first interest point retrieval unit on the first interest point retrieval layer.

And a substep S11136, performing interest point retrieval on the advertisement attention content and the second user interest learning information extracted by the first interest point retrieval unit on the first interest point retrieval layer to obtain corresponding interest point retrieval elements of the second interest point retrieval unit on the second interest point retrieval layer.

In the sub-step S11137, the advertisement attention content extracted by the first interest point retrieving unit in the first interest point retrieving layer is obtained and used as the corresponding interest point retrieving element of the first interest point retrieving unit in the second interest point retrieving layer.

And a substep S11138, performing first advertisement attention content matching of the second interest point retrieval layer on the corresponding interest point retrieval elements of the second interest point retrieval layer of the first interest point retrieval unit through the first interest point retrieval unit to obtain the advertisement attention content firstly extracted by the first interest point retrieval unit on the second interest point retrieval layer.

And a substep S111391, performing first advertisement attention content matching of a second interest point retrieval layer on the corresponding interest point retrieval elements of the second interest point retrieval unit in the second interest point retrieval layer through the second interest point retrieval unit to obtain the advertisement attention content extracted for the first time by the second interest point retrieval unit in the second interest point retrieval layer.

In the sub-step S111392, the advertisement attention content first extracted by the first interest point retrieval unit at the second interest point retrieval layer is transferred to the second interest point retrieval unit, and the advertisement attention content first extracted by the second interest point retrieval unit at the second interest point retrieval layer is transferred to the first interest point retrieval unit.

In the sub-step S111393, the first interest point retrieving unit performs the interest point retrieval on the advertisement attention content first extracted by the first interest point retrieving unit in the second interest point retrieving layer and the advertisement attention content delivered by the second interest point retrieving unit, and performs the advertisement attention content matching on the content matching information.

In the sub-step S111394, the second interest point retrieving unit performs the interest point retrieval on the advertisement attention content first extracted by the second interest point retrieving unit in the second interest point retrieving layer and the advertisement attention content delivered by the first interest point retrieving unit, and performs the advertisement attention content matching on the content matching information.

In the sub-step S111395, the interest point retrieval is performed on the advertisement interest content extracted by the first interest point retrieval unit in the second interest point retrieval layer, the advertisement interest content extracted by the second interest point retrieval unit in the second interest point retrieval layer, and the third user interest learning information, so as to obtain the interest point retrieval element corresponding to the third interest point retrieval unit in the third interest point retrieval layer.

In the sub-step S111396, the first interest point retrieving unit performs advertisement attention content matching of the third interest point retrieving layer based on the advertisement attention content extracted by the first interest point retrieving unit in the second interest point retrieving layer, the second interest point retrieving unit performs advertisement attention content matching of the third interest point retrieving layer based on the advertisement attention content extracted by the second interest point retrieving unit in the second interest point retrieving layer, and the third interest point retrieving unit performs advertisement attention content matching of the third interest point retrieving layer on the interest point retrieving element corresponding to the third interest point retrieving unit in the third interest point retrieving layer.

Further, in a possible implementation manner, for step S1114, in retrieving content based on the interest point, obtaining visitor interest data of the advertisement drainage skip subgraph under the advertisement push feedback information may be implemented by the following exemplary sub-steps, which are described in detail below.

And a substep S11141, determining a corresponding retrieval model of the advertisement push feedback information.

And a substep S11142, inputting the interest point retrieval content into a retrieval model, and obtaining visitor interest data of the advertisement drainage skip subgraph under the advertisement push feedback information through the retrieval model.

The interest point retrieval model and the configuration mode of the retrieval model can be realized by the following exemplary embodiments:

(1) and obtaining the interest point retrieval content sample, the interest point retrieval model and the retrieval model, wherein the retrieval label of the interest point retrieval content sample is used for representing the labeling result of the interest point retrieval content sample under the advertisement push feedback information.

(2) And retrieving content samples based on the interest points, and determining user interest learning information samples of a plurality of advertisement drainage paths.

(3) And respectively inputting the user interest learning information samples into a plurality of interest point retrieval units in the interest point retrieval model, and performing at least one time of advertisement attention content matching through the interest point retrieval units to obtain at least one interest point retrieval content.

The method comprises the steps that at least one time of advertisement attention content matching through an interest point retrieval unit is carried out on the basis of an associated big data retrieval partition, and the associated big data retrieval partition is associated with interest point retrieval contents extracted by other interest point retrieval units of a plurality of interest point retrieval units.

(4) And performing interest point retrieval on the advertisement attention contents output by the interest point retrieval units to obtain target interest point retrieval contents.

(5) And inputting the target interest point retrieval content into the retrieval model, and obtaining an interest point retrieval result of the interest point retrieval content sample under the advertisement push feedback information through the retrieval model.

(6) And searching a searching model and a searching model for the interest points of the advertisement strategy based on the interest point searching result and the searching labels.

Further, in step S1112, in the process of obtaining the corresponding advertisement drainage skip sub-graph based on the advertisement drainage search segment, the corresponding advertisement drainage skip sub-graph corresponding to the advertisement drainage search segment may be obtained from the preset search partition set.

On the basis, in the step, in the process of inputting the interest point retrieval content into the retrieval model and obtaining the visitor interest data of the advertisement drainage skip subgraph under the advertisement push feedback information through the retrieval model, the target interest point retrieval content can be input into the retrieval model, and the retrieval advertisement drainage search is carried out on the target interest point retrieval content through the retrieval model to obtain the visitor interest data of the target interest point retrieval content.

Further, in a possible implementation manner, regarding step S1114, in the process of performing a point of interest search on a plurality of advertisement attention contents output by a plurality of point of interest search units to obtain a point of interest search content, the following exemplary sub-steps may be implemented, which are described in detail below.

In the sub-step S11143, a plurality of advertisement attention contents output by the plurality of interest point retrieving units are obtained, and a target advertisement attention content of a global advertisement drainage path among the plurality of advertisement attention contents is determined.

And a substep S11144 of performing interest point retrieval on other advertisement attention contents except the target advertisement attention content in the plurality of advertisement attention contents, wherein the advertisement drainage path of the other advertisement attention contents after the interest point retrieval is the same as the advertisement drainage path of the target advertisement attention content.

And the substep S11145, listing other advertisement attention contents and target advertisement attention contents after the interest point retrieval to obtain the interest point retrieval contents.

Fig. 3 is a schematic diagram of functional modules of an advertisement data access device 300 based on a block chain and artificial intelligence according to an embodiment of the present disclosure, and this embodiment may divide the functional modules of the advertisement data access device 300 based on the block chain and artificial intelligence according to a method embodiment executed by the big data center 100, that is, the following functional modules corresponding to the advertisement data access device 300 based on the block chain and artificial intelligence may be used to execute various method embodiments executed by the big data center 100. The device 300 may include a calling module 310, a first extracting module 320, a second extracting module 330, and a determining module 340, and the functions of the functional modules of the device 300 are described in detail below.

The invoking module 310 is configured to invoke an advertisement push control process for pushing an advertisement through a blockchain through an advertisement data running process of the big data center, and obtain advertisement access data related to an advertisement access party based on the advertisement push control process. The invoking module 310 may be configured to execute the step S110, and the detailed implementation of the invoking module 310 may refer to the detailed description of the step S110.

The first extraction module 320 is configured to perform feature extraction on the advertisement access data to obtain advertisement access data feature information, determine push key element prediction information corresponding to a current advertisement access party according to the advertisement access data feature information, and extract predicted push element feature information of the push key element prediction information. The first extraction module 320 may be configured to perform the step S120, and the detailed implementation of the first extraction module 320 may refer to the detailed description of the step S120.

The second extraction module 330 is configured to perform feature extraction on feature information corresponding to the historical pushed element information and advertisement access data feature information to obtain reference pushed element feature information. The second extraction module 330 may be configured to perform the step S130, and the detailed implementation of the second extraction module 330 may refer to the detailed description of the step S130.

The determining module 340 is configured to fuse the predicted pushed element feature information and the reference pushed element feature information to obtain second fused feature information, perform feature analysis on the second fused feature information to obtain current pushed element information, determine a target pushed element set according to the current pushed element information and the historical pushed element information, and push corresponding advertisement data to an advertisement service after performing advertisement data access based on the target pushed element set. The determining module 340 may be configured to perform the step S140, and the detailed implementation of the determining module 340 may refer to the detailed description of the step S140.

It should be noted that the division of the modules of the above apparatus is only a logical division, and the actual implementation may be wholly or partially integrated into one physical entity, or may be physically separated. And these modules may all be implemented in software invoked by a processing element. Or may be implemented entirely in hardware. And part of the modules can be realized in the form of calling software by the processing element, and part of the modules can be realized in the form of hardware. For example, the invoking module 310 may be a separate processing element, or may be integrated into a chip of the apparatus, or may be stored in a memory of the apparatus in the form of program code, and a processing element of the apparatus invokes and executes the function of the invoking module 310. Other modules are implemented similarly. In addition, all or part of the modules can be integrated together or can be independently realized. The processing element described herein may be an integrated circuit having signal processing capabilities. In implementation, each step of the above method or each module above may be implemented by an integrated logic circuit of hardware in a processor element or an instruction in the form of software.

Fig. 4 shows a hardware structure diagram of a big data center 100 for implementing the above-mentioned advertisement data access method based on blockchain and artificial intelligence, according to an embodiment of the present disclosure, as shown in fig. 4, the big data center 100 may include a processor 110, a machine-readable storage medium 120, a bus 130, and a transceiver 140.

In a specific implementation process, at least one processor 110 executes computer-executable instructions stored in the machine-readable storage medium 120 (for example, the calling module 310, the first extraction module 320, the second extraction module 330, and the determination module 340 included in the block chain and artificial intelligence based advertisement data access apparatus 300 shown in fig. 3), so that the processor 110 may execute the block chain and artificial intelligence based advertisement data access method according to the above method embodiment, where the processor 110, the machine-readable storage medium 120, and the transceiver 140 are connected through the bus 130, and the processor 110 may be configured to control transceiving actions of the transceiver 140, so as to perform data transceiving with the aforementioned advertisement service terminal 200.

For a specific implementation process of the processor 110, reference may be made to the above-mentioned various method embodiments executed by the big data center 100, which implement principles and technical effects similar to each other, and details of this embodiment are not described herein again.

In the embodiment shown in fig. 4, it should be understood that the Processor may be a Central Processing Unit (CPU), other general purpose Processor, a Digital Signal Processor (DSP), an Application Specific Integrated Circuit (ASIC), etc. A general purpose processor may be a microprocessor or the processor may be any conventional processor or the like. The steps of a method disclosed in connection with the present invention may be embodied directly in a hardware processor, or in a combination of the hardware and software modules within the processor.

The machine-readable storage medium 120 may comprise high-speed RAM memory and may also include non-volatile storage NVM, such as at least one disk memory.

The bus 130 may be an Industry Standard Architecture (ISA) bus, a Peripheral Component Interconnect (PCI) bus, an Extended ISA (EISA) bus, or the like. The bus 130 may be divided into an address bus, a data bus, a control bus, and the like. For ease of illustration, the buses in the figures of the present application are not limited to only one bus or one type of bus.

In addition, the embodiment of the present application further provides a readable storage medium, in which computer-executable instructions are stored, and when a processor executes the computer-executable instructions, the method for accessing advertisement data based on block chains and artificial intelligence is implemented.

The foregoing description has been directed to specific embodiments of this disclosure. Other embodiments are within the scope of the following claims. In some cases, the actions or steps recited in the claims may be performed in a different order than in the embodiments and still achieve desirable results. In addition, the processes depicted in the accompanying figures do not necessarily require the particular order shown, or sequential order, to achieve desirable results. In some embodiments, multitasking and parallel processing may also be possible or may be advantageous.

Having thus described the basic concept, it will be apparent to those skilled in the art that the foregoing detailed disclosure is to be regarded as illustrative only and not as limiting the present specification. Various modifications, improvements and adaptations to the present description may occur to those skilled in the art, although not explicitly described herein. Such modifications, improvements and adaptations are proposed in the present specification and thus fall within the spirit and scope of the exemplary embodiments of the present specification.

Also, particular push elements are used in this description to describe embodiments of this description. Reference to "one embodiment," "an embodiment," and/or "some embodiments" means that a particular feature, structure, or characteristic described in connection with at least one embodiment of the specification. Therefore, it is emphasized and should be appreciated that two or more references to "an embodiment" or "one embodiment" or "an alternative embodiment" in various places throughout this specification are not necessarily all referring to the same embodiment. Furthermore, some features, structures, or characteristics of one or more embodiments of the specification may be combined as appropriate.

Moreover, those skilled in the art will appreciate that aspects of the present description may be illustrated and described in terms of several patentable species or contexts, including any new and useful combination of processes, machines, manufacture, or materials, or any new and useful improvement thereof. Accordingly, aspects of this description may be performed entirely by hardware, entirely by software (including firmware, resident software, micro-code, etc.), or by a combination of hardware and software. The above hardware or software may be referred to as "data block," module, "" engine, "" unit, "" component, "or" system. Furthermore, aspects of the present description may be represented as a computer product, including computer readable program code, embodied in one or more computer readable media.

The computer storage medium may comprise a propagated data signal with the computer program code embodied therewith, for example, on baseband or as part of a carrier wave. The propagated signal may take any of a variety of forms, including electromagnetic, optical, etc., or any suitable combination. A computer storage medium may be any computer-readable medium that can communicate, propagate, or transport a program for use by or in connection with an instruction execution system, or device. Program code located on a computer storage medium may be propagated over any suitable medium, including radio, cable, fiber optic cable, RF, or the like, or any combination of the preceding.

Computer program code required for the operation of various portions of this specification may be written in any one or more programming languages, including an object oriented programming language such as Java, Scala, Smalltalk, Eiffel, JADE, Emerald, C + +, C #, VB.NET, Python, and the like, a conventional programming language such as C, VisualBasic, Fortran2003, Perl, COBOL2002, PHP, ABAP, a dynamic programming language such as Python, Ruby, and Groovy, or other programming languages, and the like. The program code may execute entirely on the user's computer, partly on the user's computer, as a stand-alone software package, partly on the user's computer and partly on a remote computer or entirely on the remote computer or server. In the latter scenario, the remote computer may be connected to the user's computer through any network format, such as a Local Area Network (LAN) or a Wide Area Network (WAN), or the connection may be made to an external computer (for example, through the Internet), or in a cloud computing environment, or as a service, such as a software as a service (SaaS).

Additionally, the order in which the elements and sequences are processed, the use of alphanumeric characters, or the use of other designations in this specification is not intended to limit the order of the processes and methods in this specification, unless otherwise specified in the claims. While various presently contemplated embodiments of the invention have been discussed in the foregoing disclosure by way of example, it is to be understood that such detail is solely for that purpose and that the appended claims are not limited to the disclosed embodiments, but, on the contrary, are intended to cover all modifications and equivalent arrangements that are within the spirit and scope of the embodiments herein. For example, although the system components described above may be implemented by hardware devices, they may also be implemented by software-only solutions, such as installing the described system on an existing server or mobile device.

Similarly, it should be noted that in the preceding description of embodiments of the present specification, various features are sometimes grouped together in a single embodiment, figure, or description thereof for the purpose of streamlining the disclosure aiding in the understanding of one or more of the embodiments. This method of disclosure, however, is not intended to imply that more features than are expressly recited in a claim. Indeed, the embodiments may be characterized as having less than all of the features of a single embodiment disclosed above.

Finally, it should be understood that the examples in this specification are only intended to illustrate the principles of the examples in this specification. Other variations are also possible within the scope of this description. Thus, by way of example, and not limitation, alternative configurations of the embodiments of the specification can be considered consistent with the teachings of the specification. Accordingly, the embodiments of the present description are not limited to only those embodiments explicitly described and depicted herein.

Claims

1. A method for determining pushing elements based on artificial intelligence is applied to a big data center which is in communication connection with a plurality of advertisement service terminals, and comprises the following steps:

acquiring second fusion characteristic information obtained after fusion of the predicted push element characteristic information and the reference push element characteristic information, acquiring a push content service represented by each push element in the second fusion characteristic information, and acquiring content service characteristics of the push content service and source characteristic information corresponding to the previous n historical scheduling advertisement access records, wherein n is a positive integer;

and determining the current pushed element information and the historical pushed element information as a target pushed element set, and pushing corresponding advertisement data to the advertisement service terminal after accessing the advertisement data based on the target pushed element set.

2. The method for determining artificial intelligence-based push elements according to claim 1, wherein the performing decision tree mining on the content service features, the source feature information corresponding to the previous n historical scheduled advertisement access records, and the advertisement source configuration features to obtain the source feature information corresponding to the current advertisement source comprises:

3. The method for determining artificial intelligence-based push elements according to claim 1, wherein the performing decision tree mining on the source feature information and the content service feature corresponding to the current advertisement source to obtain the current push element information corresponding to the current advertisement source comprises:

4. The method according to claim 1, wherein the step of obtaining second fused feature information obtained by fusing the predicted pushed feature information and the reference pushed feature information comprises:

calling the advertisement pushing control process through the advertisement data running process of the big data center, and acquiring advertisement access data related to an advertisement access party based on the advertisement pushing control process;

fusing the predicted pushed element feature information and the reference pushed element feature information to obtain second fused feature information;

the crowd interest point information specifically comprises crowd interest point coverage labels, interest point advertisement labels and access attribute information, the crowd interest point coverage labels are used for representing crowd labels covered by the crowd interest points, the interest point advertisement labels are used for indicating advertisement attention contents of the crowd interest points, the advertisement attention contents refer to user click content areas corresponding to advertisement playing streams, and the access attribute information is used for indicating advertisement access conditions in the crowd interest points.

5. The method for determining pushing elements based on artificial intelligence according to claim 1, wherein the step of obtaining visitor interest data corresponding to advertisement pushing feedback information of a target advertisement pushing task and performing crowd interest point retrieval based on the visitor interest data comprises:

acquiring advertisement control node information corresponding to each advertisement push segment of the advertisement service terminal, and performing advertisement drainage search on advertisement push feedback information of a target advertisement push task based on the advertisement control node information to acquire a corresponding advertisement drainage search segment;

acquiring corresponding advertisement drainage jumping subgraphs based on the advertisement drainage searching fragments, and determining user interest learning information of a plurality of advertisement drainage paths based on the advertisement drainage jumping subgraphs;

respectively inputting the user interest learning information into a plurality of interest point retrieval units in an interest point retrieval model, and performing at least one time of advertisement attention content matching through the interest point retrieval units to obtain at least one advertisement attention content; the method comprises the steps that at least one time of matching of advertisement attention content through an interest point retrieval unit is carried out on the basis of an associated big data retrieval partition, the associated big data retrieval partition is associated with the advertisement attention content extracted by other interest point retrieval units of a plurality of interest point retrieval units, wherein the interest point retrieval unit is obtained on the basis of artificial intelligence network and corresponding training samples, and the training samples comprise user interest learning information samples and corresponding advertisement attention content labels;

and searching interest points of a plurality of advertisement attention contents output by the interest point searching units to obtain interest point searching contents, obtaining visitor interest data of the advertisement diversion skip subgraph under the advertisement push feedback information based on the interest point searching contents, and searching crowd interest points based on the visitor interest data of the advertisement diversion skip subgraph under the advertisement push feedback information.

6. The artificial intelligence based push element determination method of claim 5, the user interest learning information includes at least first, second and third user interest learning information, the interest point retrieval model comprises a first interest point retrieval unit, a second interest point retrieval unit and a third interest point retrieval unit, wherein the first user interest learning information, the second user interest learning information and the third user interest learning information respectively correspond to the user interest learning information of the interest coverage learning segment, the interest data source segment and the interest output source segment, the first interest point retrieval unit, the second interest point retrieval unit and the third interest point retrieval unit respectively correspond to the interest point retrieval units of the interest coverage learning segment, the interest data source segment and the interest output source segment;

the step of inputting the user interest learning information into a plurality of interest point retrieval units in an interest point retrieval model respectively, and performing at least one time of advertisement attention content matching through the interest point retrieval units to obtain at least one advertisement attention content comprises the following steps:

inputting the first user interest learning information into a first interest point retrieval unit to perform advertisement attention content matching of a first interest point retrieval layer, and obtaining advertisement attention content extracted by the first interest point retrieval unit on the first interest point retrieval layer;

performing interest point retrieval on the advertisement attention content extracted by the first interest point retrieval unit on a first interest point retrieval layer and the second user interest learning information to obtain corresponding interest point retrieval elements of the second interest point retrieval unit on a second interest point retrieval layer;

acquiring advertisement attention content extracted by the first interest point retrieval unit on a first interest point retrieval layer, and using the advertisement attention content as an interest point retrieval element corresponding to the first interest point retrieval unit on a second interest point retrieval layer;

performing first advertisement attention content matching of a second interest point retrieval layer on the corresponding interest point retrieval elements of the first interest point retrieval unit in the second interest point retrieval layer through the first interest point retrieval unit to obtain first extracted advertisement attention content of the first interest point retrieval unit in the second interest point retrieval layer;

performing first advertisement attention content matching of a second interest point retrieval layer on the corresponding interest point retrieval elements of the second interest point retrieval unit in the second interest point retrieval layer through the second interest point retrieval unit to obtain first extracted advertisement attention content of the second interest point retrieval unit in the second interest point retrieval layer;

the advertisement attention content firstly extracted by the first interest point retrieval unit at a second interest point retrieval layer is transmitted to the second interest point retrieval unit, and the advertisement attention content firstly extracted by the second interest point retrieval unit at the second interest point retrieval layer is transmitted to the first interest point retrieval unit;

the first interest point retrieval unit is used for carrying out interest point retrieval on the advertisement attention content firstly extracted by the first interest point retrieval unit in a second interest point retrieval layer and the advertisement attention content transmitted by the second interest point retrieval unit, and carrying out advertisement attention content matching on content matching information;

the second interest point retrieval unit is used for carrying out interest point retrieval on the advertisement attention content firstly extracted by the second interest point retrieval unit in a second interest point retrieval layer and the advertisement attention content transmitted by the first interest point retrieval unit, and carrying out advertisement attention content matching on content matching information;

performing interest point retrieval on the advertisement attention content extracted by the first interest point retrieval unit on a second interest point retrieval layer, the advertisement attention content extracted by the second interest point retrieval unit on the second interest point retrieval layer and the third user interest learning information to obtain an interest point retrieval element corresponding to the third interest point retrieval unit on a third interest point retrieval layer;

the first interest point retrieval unit is used for matching the advertisement attention content of a third interest point retrieval layer based on the advertisement attention content extracted by the first interest point retrieval unit on the second interest point retrieval layer, the second interest point retrieval unit is used for matching the advertisement attention content of the third interest point retrieval layer based on the advertisement attention content extracted by the second interest point retrieval unit on the second interest point retrieval layer, and the third interest point retrieval unit is used for matching the advertisement attention content of the third interest point retrieval layer with the corresponding interest point retrieval element of the third interest point retrieval unit on the third interest point retrieval layer.

7. The advertisement data access system based on the block chain and the artificial intelligence is characterized by comprising a big data center and a plurality of advertisement service terminals which are in communication connection with the big data center;

the big data center is used for:

8. A big data center, comprising a processor, a machine-readable storage medium, and a network interface, wherein the machine-readable storage medium, the network interface, and the processor are connected via a bus system, the network interface is configured to be communicatively connected to at least one advertisement service terminal, the machine-readable storage medium is configured to store a program, instructions, or code, and the processor is configured to execute the program, instructions, or code in the machine-readable storage medium to perform the artificial intelligence based pushed element determination method of any one of claims 1-6.